近红外光谱法快速无损测定奶粉的脂肪含量  

Rapid nondestructive determination of milk power fat content by near-infrared spectroscopy(NIR)

作　　者：何佳艳[1] 李亭 郭长凯 胡蝶 邹婷婷[1] 王莹 

机构地区：[1]北京工商大学北京市食品风味化学重点实验室/食品添加剂与配料北京高校工程研究中心,北京100048 [2]吉林省食品检验所,吉林长春130022

出　　处：《食品与发酵工业》2017年第10期228-233,共6页Food and Fermentation Industries

摘　　要：将11种奶粉原样配制及组成77个奶粉样本,以脂肪含量为检测指标,结合偏最小二乘法开展近红外光谱定量分析研究。2次异常光谱剔除,识别出异常样本(14,52,76)并予以剔除。74个奶粉样本进行平滑、导数和标准变量变换等6种光谱预处理,确定标准正态变量变换结合Norris一阶导为最佳光谱预处理方式,其交叉验证均方根误差为0.354 7,交叉验证相关系数平方达到0.990 8;最佳前处理光谱结合3种波段选择方法优化模型性能,与全光谱模型形成对比,确定随机蛙跳(random frog,RF)为最佳波段选择方式,其模型的训练集和测试集相关系数平方分别为0.997 2和0.997 0,训练集和测试集均方根误差分别为0.186 2和0.198 2。结果表明:采用蒙特卡罗异常光谱剔除(Monte-Carlo sampling,MCS),光谱预处理结合随机蛙跳波段优化技术可提高奶粉脂肪近红外定量模型的泛化性和预测能力。Eleven milk powers were diluted into 77 samples. Choosing fat content as the index, the near-infrared spectrum quantitative analysis combined with the partial least squares method were used. Abnormal samples （14, 52 and 76） shall be recognized and removed through twice abnormal spectrum removals. Seventy four milk power samples were studied by six spectrum pretreatments such as smoothing, derivative and standard variable transformation. The best pretreatment method was the standard normal variable transformation combined with Norris first order derivative, which root-mean-square error of cross validation was 0. 354 7 and correlation coefficient square of cross validation reached 0. 990 8. The established model performance improved by optimal pretreatment spectrum with three band se- lections. Random Frog （RF） was the optimal band selection method after comparing with the full-spectrum model, which correlation coefficient square of training set and test set were 0. 997 2 and 0. 997 0 respectively. The root-mean- square of training set and test set were 0. 186 2 and 0. 198 2 respectively. The result is that Monte-Carlo sampling （MCS）, spectrum pretreatment and band optimization technology can improve the generalization and prediction in milk power fat dectection near-infrared quantitative model.

关 键 词：近红外光谱技术 无损检测 脂肪 蒙特卡罗异常光谱剔除(Monte-Carlo sampling MCS) 随机蛙跳(random frog RF) 

分 类 号：O657.33[理学—分析化学]